Blockchains are integrated with an emerging range of decentralized AI networks. Photo by Clarisse Croset on Unsplash.
Key Takeaways
- Decentralized AI offers an alternative to the centralized solutions that currently dominate.
- Blockchains are used to provide tokenomic systems for decentralized AI networks.
- Platforms like Aethir and LibertAI have deployed their own AI tokens that incentivize participation and can be used pay for services.
In recent years, the intersection of AI and blockchain has emerged as one of the most exciting areas of technological innovation. Both fields hold transformative potential, and where they meet, they promise to change how we interact with digital technology forever.
From training to inference, blockchains can help decentralize every stage of the AI development and deployment cycle. But why is there such a need for decentralized AI in the first place?
Privacy and Decentralization
In the Web2 era, companies like Google and Meta built empires on top of user data, often at the expense of privacy and digital autonomy.
According to Jonathan Schemoul , who founded the decentralized AI platform LibertAI , if AI services follow their current trajectory, they are on course to follow the pattern established by search and social media.
“People should be concerned because they are giving a lot of data to these chatbots,” he observed. As adoption increases and people integrate AI assistants into more areas of their lives, he said the amount of information centralized parties will be able to collect could eclipse the massive data troves already compiled by Big Tech digital platforms.
To solve this problem and give users greater control over their personal data, platforms like LibertAI are building decentralized networks that eliminate the need for centralized servers.
DePIN for AI
Modern Decentralized Physical Infrastructure Networks (DePINs) have roots in the first blockchains, although the technologies differ in some important ways.
Startups like Render, NodeAI and Aethir are using DePINs to meet the rising demand for GPUs for AI training. For developers who need dynamic GPU access, such solutions are significantly more flexible than traditional cloud providers, which typically lock customers into rigid contracts based on their expected needs.
Meanwhile, LibertAI and its peers are focused on CPU provisioning, which is needed for AI inference rather than training.
Whether GPU- or CPU-based, DePins often utilize blockchains to manage and record ownership and a unique tokenomic model is starting to emerge.
Decentralized AI tokenomics
To incentivize participation, decentralized AI networks typically reward nodes with tokens, and AI coins have emerged as one of the most exciting crypto narratives of 2024.
AI tokens like Render’s RNDR and Aethir’s ATH have an important transactional utility, functioning as the default currency with which users pay node operators for compute. But simply distributing a bunch of tokens to participants without any thought to long-term sustainability, poses the risk that nodes will disconnect after the bubble bursts.
“We use the token very much as a stabilizing force,” Aether CEO Mark Rydon told CCN.
Rather than rushing to scale by promoting short-term gains, “we’ve really focused on […] having incentives for people not just to join, but also to stick around long-term.”
While RNDR and ATH are both deployed as ERC-20 tokens on Ethereum, LibertAI took a different path.
Launched on Base on Tuesday, June 18, LTAI benefits from lower transaction fees compared Layer 1 Ethereum.
“We needed a chain where it would be cheap to do token streaming and legal payments which are needed for the API. And on the other hand, we wanted to be close to the Ethereum ecosystem and have a lot of users already there,” Schemoul explained. “Base actually checked all these marks, and for institutional adoption, Base is a good choice as well,” he added.
